AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Generative Well-intentioned Networks

Justin Cosentino, Jun Zhu

NeurIPS 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Learning from unlabeled and positive-unlabeled data · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Transform and reclassify rejected observations with WGWIN-GPBayesian LeNet-5 · MNIST Digits test, rejected subset (τ=0.50–0.99)accuracy: 35.36±8.66; 30.59±2.64; 27.39±2.03; 18.16±2.55; 15.86±2.07; 9.02±0.94Improved
Transform and reclassify rejected observations with WGWIN-GPBayesian LeNet-5 · MNIST Fashion test, rejected subset (τ=0.50–0.99)accuracy: 18.91±3.61; 14.87±1.78; 11.29±0.87; 9.04±0.83; 7.14±0.61; 4.79±0.34Improved
Transform and reclassify rejected observations with WGWIN-GPImproved Bayesian neural network · MNIST Digits test, rejected subset (τ=0.70–0.99)accuracy: 12.50±14.17; 9.01±7.81; 8.36±8.58; 7.58±6.97; 2.62±3.80Improved
Transform and reclassify rejected observations with WGWIN-GPImproved Bayesian neural network · MNIST Fashion test, rejected subset (τ=0.50–0.99)accuracy: 9.42±11.05; 10.95±2.35; 9.67±2.93; 8.53±1.91; 6.15±2.11; 3.09±1.59Improved
Use a higher certainty rejection thresholdBNN+GWIN · MNIST Digits and MNIST Fashion test, rejected subsetaccuracy gain over BNN: MNIST Digits: 35.36±8.66 to 9.02±0.94; MNIST Fashion: 18.91±3.61 to 4.79±0.34Mixed
Use the mean rather than median as the certainty metricWGWIN-GP with Bayesian neural network · MNIST experimentsGWIN performanceNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy